Millennial scale dynamics of southern amazonian rain forests


Published on

Published in: Technology
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Millennial scale dynamics of southern amazonian rain forests

  1. 1. DOI: 10.1126/science.290.5500.2291, 2291 (2000);290Science, et al.Francis E. MayleMillennial-Scale Dynamics of Southern Amazonian Rain ForestsThis copy is for your personal, non-commercial use only.clicking here.colleagues, clients, or customers by, you can order high-quality copies for yourIf you wish to distribute this article to othershere.following the guidelinescan be obtained byPermission to republish or repurpose articles or portions of articles):March 23, (this infomation is current as ofThe following resources related to this article are available online at of this article at:including high-resolution figures, can be found in the onlineUpdated information and services, be found at:Supporting Online Material at:can berelated to this articleA list of selected additional articles on the Science Web sites, 4 of which can be accessed free:cites 21 articlesThis article115 article(s) on the ISI Web of Sciencecited byThis article has been articles hosted by HighWire Press; see:cited byThis article has been collections:This article appears in the followingregistered trademark of aScience2000 by the American Association for the Advancement of Science; all rights reserved. The titleCopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005.(print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScienceonMarch23,2011www.sciencemag.orgDownloadedfrom
  2. 2. ing the mid-Pliocene to feed the enhancedproduction of NADW (3, 5, 6, 38). However,our results suggest that the manifestation ofthe BOC as a characteristic intrusion of coldsurface water across the modern South Atlan-tic (Fig. 1) would have been less pronouncedduring the mid-Pliocene, because the BCwould have upwelled less of the cool SACW.Our mid-Pliocene SST estimates for the BCupwelling system (ϳ26°C) are similar to theaverage annual SST for the modern oligotro-phic waters of the western South Atlantic(ϳ25°C) at the same latitude (Fig. 1). How-ever, global mid-Pliocene warmth wouldhave had a basinwide influence, so that theSST for the western boundary would proba-bly have remained higher than for the easternboundary (12), because our paleoproductivityrecords and previous studies indicate that adegree of upwelling continued throughout themid-Pliocene (18–20).The processes related to enhanced BCupwelling discussed here may have occurredat the eastern boundaries of the other threemajor Atlantic and Pacific Ocean basins.These systems could have provided addition-al long-term sinks for atmospheric CO2through the hypothetical mechanism outlinedabove. The cooling of the Canary Current(24, 25) suggests that the process may haveoccurred at the North Atlantic’s easternboundary. The development of the U37KЈtech-nique for quantitative late Neogene SST re-construction in sediment cores from a varietyof oceanographic settings should allow theseand other hypotheses about changes in globaland regional paleotemperatures over the past5 million years to be tested further.References and Notes1. D. A. Hodell, K. Venz, in Antarctic Research Series, J. P.Kennett, D. A. Warnke, Eds. (American GeophysicalUnion, Washington, DC, 1992), vol. 56, pp. 265–310.2. N. J. Shackleton, M. A. Hall, D. Pate, Proc. ODP Sci.Res. 138, 337 (1995).3. G. H. Haug, R. Tiedemann, Nature 393, 673 (1998).4. C. France-Lanord, L. A. Derry, Nature 390, 65 (1997).5. K. Billups, A. C. Ravelo, J. C. Zachos, Paleoceanogra-phy 13, 84 (1998).6. W. H. Berger, G. Wefer, in The South Atlantic: Presentand Past Circulation, G. Wefer, W. H. Berger, G.Siedler, D. J. Webb, Eds. (Springer, Heidelberg, Ger-many, 1996), pp. 363-410.7. K. G. Miller, R. G. Fairbanks, G. S. Mountain, Pale-oceanography 2, 1 (1987).8. J. A. Barron, J. G Baldauf, in Productivity of the Ocean:Present and Past, W. H. Berger, V. S. Smetacek, G.Wefer, Eds. (Wiley, New York, 1989), pp. 341–354.9. W. F. Ruddiman, Paleoceanography 4, 353 (1989).10. D. Rind, J. Geophys. Res. 103, 5943 (1998).11. L. C. Sloan, T. J. Crowley, D. Pollard, Mar. Micropal-eontol. 27, 51 (1996).12. H. J. Dowsett, J. Barron, R. Poore, Mar. Micropaleon-tol, 27, 13 (1996).13. H. J. Dowsett, R. Z. Poore, Mar. Micropaleontol. 16, 1(1990).14. The U37KЈ index is based on the ratio of di- andtri-unsaturated n-C37 alkenones produced by specificspecies of haptophyte algae.15. S. C. Brassell, G. Eglinton, U. Pflaumann, M. Sarnthein,Nature 320, 129 (1986).16. J. R. E. Lutjeharms, P. L. Stockton, S. Afr. J. Mar. Sci. 5,35 (1987).17. S. Levitus, T. Boyer, World Ocean Atlas 1994. Vol. 4,Temperature (U.S. Department of Commerce, Wash-ington, DC, 1994).18. L. Diester-Haass, P. A. Meyers, P. Rothe, in UpwellingSystems: Evolution since the Early Miocene, C. P.Summerhayes, W. L. Prell, K. C. Emeis, Eds. (Geolog-ical Society Special Publication No. 63, London,1992), pp. 331–342.19. H. Dowsett, D. Willard, Mar. Micropaleontol. 27, 181(1996).20. W. W. Hay, J. C. Brock, in Upwelling Systems: Evolu-tion since the Early Miocene, C. P. Summerhayes,W. L. Prell, K. C. Emeis, Eds. (Geological SocietySpecial Publication No. 63, London, 1992), pp. 463–497.21. W. H. Berger et al., Proc. ODP Init. Rep. 175, 505(1998).22. R. Tiedemann, M. Sarnthein, N. J. Shackleton, Pale-oceanography 9, 619 (1994).23. P. J. Mu¨ller, M. Cepek, G. Ruhland, R. R. Schneider,Palaeogeogr. Palaeoclimatol. Palaeoecol. 135, 71(1997).24. T. D. Herbert, J. D. Schuffert, Proc. ODP Sci. Res.159T, 17 (1998).25. U. Pflaumann, M. Sarnthein, K. Ficken, A. Grothmann,A. Winkler, Proc. ODP Sci. Res. 159T, 3 (1998).26. I. T. Marlowe, S. C. Brassell, G. Eglinton, J. C. Green,Chem. Geol. 88, 349 (1990).27. T. D. Herbert et al., Paleoceanography 13, 263(1998).28. P. J. Mu¨ller, G. Kirst, G. Ruhland, I. von Storch, A.Rosell-Mele´, Geochim. Cosmochim. Acta 62, 1757(1998).29. K. -C. Emeis, H. Doose, A. Mix, D. Schulz-Bull, Proc.ODP Sci. Res. 138, 605 (1995).30. M. J. L. Hoefs, G. J. M. Versteegh, W. I. C. Rijpstra, J. Leeuw, J. S. Sinninghe Damste´, Paleoceanography13, 42 (1998).31. C. R. Gong, D. J. Hollander, Geochim. Cosmochim.Acta 63, 405 (1999).32. C. B. Lange, W. H. Berger, H. -L. Lin, G. Wefer,Shipboard Scientific Party, Mar. Geol. 161, 93 (1999).33. U. F. Treppke et al., J. Mar. Res. 54, 991 (1996).34. A. Abelmann, R. Gersonde, V. Speiss, in GeologicalHistory of the Polar Oceans: Arctic Versus Antarctic,U. Bleil, J. Thiede, Eds. (Kluwer, Dordrecht, Nether-lands, 1990), pp. 729–759.35. W. F. Ruddiman, T. R. Janecek, Proc. ODP Sci. Res.108, 211 (1989).36. S. A. Hovan, D. K. Rea, Proc. ODP Sci. Res. 121, 219(1991).37. G. H. Haug, D. M. Sigman, R. Tiedemann, T. F. Peder-son, M. Sarnthein, Nature 401, 779 (1999).38. M. E. Raymo, B. Grant, M. Horowitz, G. H. Rau, Mar.Micropaleontol. 27, 313 (1996).39. A. J. Watson, in Upwelling in the Ocean: ModernProcesses and Ancient Records, C. P. Summerhayes,K.-C. Emeis, M. V. Angel, R. L. Smith, B. Zeitzschel, Eds.(Wiley, New York, 1995), pp. 321–336.40. P. B. de Menocal, Science 270, 53 (1995)41. K E. Reed, J. Hum. Evol. 32, 289 (1997).42. A. Rosell-Mele´, thesis, University of Bristol, UK(1994).43. Upwelling species are associated with colder up-welled waters of the high-productivity coastal band(such as Chaetoceros spp., mainly C. radicans and C.cinctus; and Thalassionema nitzschioides var. nitz-schioides), SO indicates a Southern Ocean assem-blage composed of the subantarctic diatoms Pro-boscia barboi and Thalassiothrix antarctica and of theantarctic/subantarctic radiolarian Cycladophora plio-cenica; “warm” indicates a broad variety of specieswidely distributed in the tropical and subtropicalAtlantic [such as Alveus (ϭ Nitzschia) marinus, Azpei-tia nodulifera, A. africana, Fragilariopsis (ϭ Pseu-doeunotia) doliolus, Hemidiscus cuneiformis, Rhizoso-lenia bergonii, Roperia tesselata, Stellarima stellaris,and Thalassionema nitzschioides var. parva]; and“neritic” indicates planktonic (Actinocyclus spp.,Coscinodiscus centralis, C. gigas, C. radiatus, Delphi-neis karstenii, Stephanopyxis spp., and Thalassiosiraangulata) and tycopelagic and benthic species (suchas Actinoptychus senarius, A. vulgaris, Cocconeis sp.,Delphineis surirella, and Paralia sulcata) characteristicof nearshore waters.44. Shipboard Scientific Party, Proc. ODP Init. Rep. 175,339 (1998).45. We thank W. Berger, K. Emeis, P. Farrimond, and R.Schneider for discussions and the ODP for providingsamples. Supported by a studentship from the UKNatural Environment Research Council to J.R.M. andby funds from Joint Oceanographic Institutions, Inc.,and the U.S. Science Support Program to C.B.L., fromDeutsche Forschungsgemeinschaft to G.W., and fromthe European Union Environment and Climate Pro-gram to A.R.-M.4 August 2000; accepted 16 November 2000Millennial-Scale Dynamics ofSouthern Amazonian RainForestsFrancis E. Mayle,1* Rachel Burbridge,1Timothy J. Killeen2,3Amazonian rain forest–savanna boundaries are highly sensitive to climaticchange and may also play an important role in rain forest speciation. However,their dynamics over millennial time scales are poorly understood. Here, wepresent late Quaternary pollen records from the southern margin of Amazonia,which show that the humid evergreen rain forests of eastern Bolivia have beenexpanding southward over the past 3000 years and that their present-day limitrepresents the southernmost extent of Amazonian rain forest over at least thepast 50,000 years. This rain forest expansion is attributed to increased seasonallatitudinal migration of the Intertropical Convergence Zone, which can in turnbe explained by Milankovitch astronomic forcing.Understanding the long-term dynamics ofAmazonian rain forest–savanna boundariesover millennial time scales can provide im-portant insights into Amazonian paleocli-mates and may also improve understandingof rain forest biodiversity (1). However, thelate Quaternary history of forest-savannadynamics of southern Amazonia is poorlyunderstood, based predominantly on con-troversial pollen data from two sites. OneR E P O R T SCIENCE VOL 290 22 DECEMBER 2000 2291onMarch23,2011www.sciencemag.orgDownloadedfrom
  3. 3. of these sites, Carajas, is a lake on top of a700-m plateau (inselberg) in Para´ State,Brazil (southeast Amazonia), which is cov-ered by edaphically controlled savanna.The assertion (2–4) that the pollen recordfrom this site reflects regional vegetationchanges in the Amazon lowlands ratherthan local vegetation changes on the pla-teau is controversial (5). The paleoecologi-cal importance of pollen data from the sec-ond site, a valley fill in Katira Creek, Ron-doˆnia, Brazil (southwest Amazonia) (6, 7),has also proved to be contentious (5), withdisagreement over whether the pollenrecord (compiled from only five pollenspectra) reflects regional vegetation historyor merely local catchment changes. Al-though analyses of soil carbon isotopeshave revealed late Quaternary fluctuationsin southern Amazonian forest-savannaboundaries (8–10), determination of thespecific kinds of forest communities thatexisted in the past (e.g., evergreen rainforest versus semideciduous dry forest) hasnot been possible from carbon isotope dataalone. Here, we present paleoecological ev-idence for rain forest–savanna dynamics ofsouthern Amazonia, quantified both spa-tially and temporally, spanning the past50,000 years.The study area is Noel Kempff MercadoNational Park (NKMNP) (Fig. 1, A and B),a protected area of 1.5 million hectares thatis located on the Precambrian Shield ofeastern Bolivia (11), at the southern marginof Amazonia, encompassing an ecotone be-tween humid evergreen rain forest and drysemideciduous forests and savannas (12)(Fig. 1, A and B). The region has a stronglyseasonal climate produced by latitudinalshifts of the Intertropical ConvergenceZone (ITCZ) (13, 14) (Fig. 1A), which isthe overriding control over the geographi-cal position of the southern limit of Ama-zonian rain forest, presently located at thesouthern border of NKMNP (15°S).Laguna Bella Vista (13°37ЈS, 61°33ЈW)(15) is ϳ120 km north of the southern limitof humid evergreen Amazonian forest com-munities. A 3-m-long sediment core wascollected in 1995 from a floating platformtoward the center of the lake (2.5-m waterdepth) by using a modified Livingstonepiston corer (16), together with a Perspexplastic tube to collect the flocculent surfacesediments. Sediment samples (1 cm3) wereprocessed by standard methods (17). Fossilpollen was identified by using a referencepollen collection of ϳ1000 taxa collectedfrom herbarium material obtained fromNKMNP. A chronological framework forthe sedimentary sequence was provided by15 accelerator mass spectrometer (AMS)radiocarbon dates (Fig. 2A) [see Web table1 (18)]. Radiocarbon dates of Holocene agewere calibrated into calendar years beforethe present (cal yr B.P.) (19). The sedimentrecord spans at least the past 50,000 years,although no sedimentation occurred at thissite during the Last Glacial Maximum(LGM) (ϳ21,000 cal yr B.P.).The high percentages (ϳ60%) of Mora-ceae pollen and low percentages of grass(Poaceae) pollen (Ͻ10%) in the surfacesediment of Laguna Bella Vista provide asignature of the rain forest surrounding thelake today (20) (Figs. 1B and 2A). ThePleistocene (ice-age) pollen spectra1Department of Geography, University of Leicester,Leicester LE1 7RH, UK. 2Center for Applied Biodiver-sity Science, Conservation International, 2501 MStreet, NW, Suite 200, Washington, DC 20037, USA.3Museo de Historia Natural “Noel Kempff Mercado,”Avenida Irala 565, Casilla 2489, Santa Cruz de laSierra, Santa Cruz, Bolivia.*To whom correspondence should be addressed. E-mail: 1. (A) Location ofstudy area in the con-text of Amazonia andSouth American cli-mate. The region hasa seasonal climate(ϳ1400 to 1500 mmof mean annual pre-cipitation, 25° to 26°Cmean annual tempera-ture), influenced bythree distinct climatesystems (13, 14). TheITCZ reaches NKMNPin the austral summer,bringing warm, moistAmazonian air massesresponsible for therainy season. Duringthe austral winter, theITCZ moves northwardand is replaced by drynortherly winds, origi-nating from the anti-cyclonic circulation ofthe South Atlantictrade winds, which areresponsible for the dryseason. Furthermore,winter temperatures occasionally drop as low as 5°C because of northernadvections of cold polar fronts originating from the South Pacific Anticycloneover Patagonia. (B) Map showing the locations of the two study sites, LagunaBella Vista and Laguna Chaplin, and the principal vegetation types withinNKMNP. The Huanchaca Plateau (600 to 900 m above sea level), whichoccupies the eastern half of NKMNP, is composed of Precambrian sandstoneand quartzite rocks of the Brazilian Shield (11) and is dominated by edaphicallyderived upland savannas (12). The adjacent lowland peneplain to the west(200 to 250 m above sea level) is blanketed by Tertiary alluvial sediments (11) and contains examples of the four other ecosystems. On well-drainedlandscapes, Amazonian rain forests predominate in the north, whereas Chiquitano dry forest communities predominate in the south; inundated forests andsavannas occupy extensive floodplains throughout the region (12).R E P O R T S22 DECEMBER 2000 VOL 290 SCIENCE www.sciencemag.org2292onMarch23,2011www.sciencemag.orgDownloadedfrom
  4. 4. [ϳ44,000 to 38,600 radiocarbon years be-fore the present (14C yr B.P.)] are dominat-ed by Alchornea, Leguminosae (Papilion-oideae), and Talisia-type pollen. They areindicative of plant communities that arevery different from those of the Holocene(possibly semideciduous dry forests). Allthree taxa include species present in theAmazonian rain forest, Cerrado (uplandgrassland, savanna, and woodland), andChiquitano dry forest biotas. However, theabsence or negligible abundance of Mora-ceae pollen show that these forest commu-nities were not rain forest.Most of the Holocene sequence is char-acterized by low percentages of Moraceaepollen (Ͻ25%) and peaks in pollen ofPoaceae (40%), Curatella americana, Mau-ritia/Mauritiella, and Isoetes. Grasses dom-inate both well-drained and inundated sa-vannas (21). The high percentages of grasspollen are typical of modern pollen spectrafrom savannas of the Brazilian highlandssouth of Amazonia (22). Preliminary char-coal data from the same core (23) revealthat charcoal and grass pollen fluctuationsare closely in phase with one another, sig-nifying that this Holocene grass pollenpeak is attributable to savanna grasses(which are subject to frequent fire) ratherthan to aquatic or shoreline grasses. Cura-tella americana is a small tree that domi-nates well-drained savanna plant communi-ties on lowland landscapes. In NKMNP,this species is typically associated withnonflooded microhabitats (e.g., termitemounds) within seasonally inundated sa-vanna landscapes; it is essentially absentfrom the upland savannas of the HuanchacaPlateau (12). The Mauritia/Mauritiella pol-len can be attributed to two closely relatedpalm taxa, Mauritia speciosa and Mauri-tiella armata, that form large mixed colo-nies characteristic of seasonally floodedmarshland (12). Isoetes is an aquatic herbrestricted to lake-margin shallows. Peaks ofthese palms and aquatics are consistentwith lake levels lower than present-day lev-els, facilitating the expansion of Isoetespopulations over central portions of thelake and the development of marshlandaround a shrinking lake margin. Consideredas a whole, this Holocene pollen assem-blage, before 2790 cal yr B.P., is indicativeof a climate drier than today’s climate, withlowered lake levels and a landscape domi-nated by seasonally inundated savannas,with microrelief provided by termitemounds colonized by C. americana andlevees supporting Moraceae-dominatedgallery forests lining the nearby river sys-tems. Evidence that the gallery forests sup-ported at least some species of the Chiqui-Fig. 2. (A) Laguna Bella Vista summary pollen percentage diagram. Per-centages are expressed as a proportion of the total land pollen sum (atleast 300 grains), including terrestrial fern spores; all aquatic taxa (e.g.,Isoetes and Pediastrum) have been excluded from the pollen sum. Only themost abundant taxa from the full complement of 263 pollen types areshown. Dots on the curves denote Ͻ0.5% presence. Analysis of pollenconcentration data shows that these percentage changes are notstatistical artifacts, but represent real vegetation change. Percent or-ganic content was determined by loss on ignition at 550°C. On thedepth scale, 0 cm represents the interface between sediment and water.The pollen record was produced by Mayle [(23); see supplementalmaterial for a more detailed pollen diagram (18)]. (B) Laguna Chaplinsummary pollen percentage diagram. Percentages were calculated inthe same manner as for (A). Only the most abundant taxa from thefull complement of 290 pollen types are shown. Comparison withpollen concentration data reveals that the percentage peak in Mau-ritia/Mauritiella pollen at the base of the record is a statistical artifactproduced by low concentrations of other pollen taxa at this level. Thepollen record was produced by Burbridge [(26); see supplementalmaterial for a more detailed pollen diagram (18)].R E P O R T SCIENCE VOL 290 22 DECEMBER 2000 2293onMarch23,2011www.sciencemag.orgDownloadedfrom
  5. 5. tano dry forest is provided by the peak inpollen of Astronium fraxinifolium, a specieswith a present-day distribution concurrentwith dry-forest formations (24). The abun-dance of Moraceae pollen increases steeplyfrom 25% to current levels of 60% between2740 Ϯ 50 14C yr B.P. (2790 cal yr B.P.)and 1650 Ϯ 40 14C yr B.P. (1530 cal yrB.P.), showing that rain forest has expand-ed into northern NKMNP only within thepast three millennia.To determine the regional importance ofthe vegetation record from Laguna BellaVista, we cored another site 100 km farthersouth. This second site, Laguna Chaplin(14°28ЈS, 61°04ЈW) (15), was cored in1998 with a hammer-driven modified Liv-ingstone piston corer (25). The morphom-etry of this basin is very similar to that ofLaguna Bella Vista, and the lake is alsosurrounded by humid evergreen Amazonianforest (Fig. 1B). The sediments were depos-ited continuously over the past 43,000years, including the LGM; chronologicalcontrol was provided by 14 AMS 14C dates(Fig. 2B) [see Web table 2 (18)]. Uniformlyhigh grass pollen percentages of 40%, to-gether with peaks in C. americana andcharcoal (26), indicate that savanna com-munities dominated the catchment of thissite continuously from ϳ40,000 to 224014C yr B.P. The Amazonian rain forestcommunities surrounding Laguna Chaplinare even younger than those around LagunaBella Vista, having become establishedonly between 2240 Ϯ 40 14C yr B.P. (2240cal yr B.P.) and 710 Ϯ 50 14C yr B.P. (660cal yr B.P.), confirming that Amazonianspecies have expanded their distributionsouthward in NKMNP in very recent times.This finding is supported by a mappingstudy (using LANDSAT satellite imagery)(12) that places the southernmost boundaryof Amazonian rain forest just 20 km southof Laguna Chaplin (Fig. 1B).Our findings show that Amazonian rainforest communities have only expandedinto NKMNP within the past three millen-nia to reach their current geographical limitat 15°S. Furthermore, our data show thatthe present-day rain forest boundary ineastern Bolivia constitutes the southern-most extent of Amazonian rain forest inSouth America over at least the past 50,000years and that, over most of this period, theITCZ must have been located north of east-ern Bolivia in austral summer, exerting aweaker influence over this region than to-day, resulting in a longer dry season.Our evidence for early to middle Holo-cene savanna in Bolivian Amazonia corre-lates with stable carbon isotope evidencefor savanna expansion between 9000 and3000 14C yr B.P. in Rondoˆnia and Amazo-nas states (western Brazilian Amazonia)(8–10) and charcoal evidence for increasedfire frequencies between 7000 and 300014C yr B.P. in Para´ State (eastern BrazilianAmazonia) (4, 27). This correlation showsthat Holocene climate aridity did not justaffect the vegetation at the southern marginof Amazonia, it affected more central andeastern parts of the Amazon basin as well.Our pollen evidence for increased pre-cipitation after ϳ2740 14C yr B.P. is furthercorroborated by a rise in the water level ofLake Titicaca between 3600 and 3200 14Cyr B.P. (reaching near-modern levels by2100 14C yr B.P.) (28–30) and increasedsnow accumulation on Sajama Mountain atϳ3000 14C yr B.P. (31), regions in theBolivian Andes that are also influenced bythe ITCZ. This recent southward shift ofthe ITCZ can be explained by orbital forc-ing according to the Milankovitch theory,which predicts a minimum in SouthernHemisphere summer insolation fromϳ12,000 to 9000 cal yr B.P., with an in-crease toward the present (13, 32).References and Notes1. T. B. Smith, R. K. Wayne, D. J. Girman, M. W. Bruford,Science 276, 1855 (1997).2. F. Soubies, K. Suguio, L. Martin, Bol. Inst. Geol. Univ.Sao Paulo Publ. Esp. 8, 233 (1991).3. M. L. Absy et al., C. R. Acad. Sci. Paris Ser. II 312, 673(1991).4. B. Turcq et al., Ambio 27, 139 (1998).5. P. A. Colinvaux, P. E. De Oliveira, M. B. Bush, Quat. Sci.Rev. 19, 141 (2000).6. M. L. Absy, T. Van der Hammen, Acta Amazonica 6(no. 3), 293 (1976).7. T. Van der Hammen, M. L. Absy, Palaeogeogr. Palaeo-climatol. Palaeoecol. 109, 247 (1994).8. L. C. R. Pessenda et al., Holocene 8, 599 (1998).9. L. C. R. Pessenda et al., Radiocarbon 40, 1013(1998).10. H. A. de Freitas et al., Quat. Res., in press.11. M. Litherland, G. Power, J. S. Am. Earth Sci. 2, 1(1989).12. T. Killeen, T. Schulenberg, Eds., The Huanchaca Pla-teau and Noel Kempff Mercado National Park, RapidAssessment Program working papers (ConservationInternational, Washington, DC, 1998).13. E. M. Latrubesse, C. G. Ramonell, Quat. Int. 21, 163(1994).14. L. Martin et al., Quat. Res. 47, 117 (1997).15. Both sites are large (4 to 6 km in diameter),shallow, flat-bottomed lakes, formed by subsi-dence of the underlying rocks along fault lines ofthe Precambrian Brazilian Shield. Laguna Bella Vis-ta is fed by small ephemeral streams in the rainyseason, which drain a small nearby quartzite ridge;it has a single, small outflowing stream that drainsinto the Rı´o Ite´nez (the Brazilian border), 6 km tothe north. Laguna Chaplin is fed by a small inflow-ing stream from the east and has a small outflow-ing stream, flowing into the Rı´o Paragua´, 15 km tothe west. Both rivers are “black water” and havetheir sources on the Brazilian Shield.16. H. E. Wright Jr., J. Sediment. Petrol. 27, 957 (1967).17. K. Faegri, J. Iversen, Textbook of Pollen Analysis(Munksgaard, Copenhagen, 1975).18. Supplemental material is available at Science Online( M. Stuiver et al., Radiocarbon 40, 1041 (1998).20. Detailed botanical inventories (T. J. Killeen, unpub-lished data) of representative 500 m by 20 m plotsof each forest ecosystem were undertaken [em-ploying standardized methods that record treesthat are Ͼ10 cm dbh (diameter at breast height)and lianas that are Ͼ2.5 cm dbh]. A 500-m line-intercept method was employed for the quantita-tive surveys of the grassland ecosystems. Theseinventories, in combination with LANDSAT satel-lite imagery, have shown that both lakes are pre-dominantly surrounded by Amazonian rain forest(humid upland and inundated/riverine evergreenforests). Both forest types are species-rich in theMoraceae family, with the genera Brosimum (6spp.), Ficus (13 spp.), Pseudolmedia (4 spp.),Cecropia (4 spp.), Sorocea (3 spp.), Pourouma (3spp.), Maquira, Helicostylis, Maclura, Naucleopsis,and Perebea. Furthermore, these forests are alsocharacterized by a high density of trees of theMoraceae family; inventories have shown that thecanopy tree Pseudolmedia laevis is the most com-mon species, representing up to 30% of all stemsin mature forest. Although Moraceae pollen cannotbe reliably distinguished from Urticaceae pollen,we are confident that most of these pollen grainsbelong to the Moraceae family, given the domi-nance of this family in the rain forests surroundingthis site, together with the fact that Urticaceae areneither diverse nor abundant in any of the fivemajor ecosystems in the region, with a total ofonly four species known for NKMNP. Furthermore,there are only two species of the Moraceae familyknown to occur in the cerrado savannas of NKMNP(Brosimum gaudichaudii and Cecropia distachya).Moraceae species are more common in the Chi-quitano dry forest, but they are not abundant,being typically restricted to the more humid valleybottoms. As such, we infer that high percentages ofthe Moraceae/Urticaceae pollen type (e.g., 40 to60%) are indicative of humid evergreen rain forestssuch as those surrounding the lake today. BecauseMoraceae is a wind-pollinated family (notwith-standing Ficus), low percentages of its pollen (e.g.,Ͻ20%) probably reflect either long-distance trans-port from distant populations or, more likely, lo-calized populations within gallery (riparian) forestsbordering the Rı´o Paragua´ and/or the Rı´o Ite´nez(Fig. 1B).21. T. J. Killeen, P. N. Hinz, J. Trop. Ecol. 8, 389 (1992).22. K. R. Ferraz-Vicentini, M. L. Salgado-Labouriau, J. S.Am. Earth Sci. 9, 207 (1996).23. F. E. Mayle, unpublished data.24. T. J. Killeen, A. Jardim, F. Mamini, N. Rojas, P. Saravia,J. Trop. Ecol. 14, 803 (1998).25. P. Colinvaux, P. E. De Oliveira, J. E. Moreno Patino,Amazon Pollen Manual and Atlas (Harwood Academ-ic, Amsterdam, 1999).26. R. Burbridge, unpublished data.27. F. Soubies, Cah. ORSTOM Ser. Geol. 1, 133 (1979–80).28. D. Wirrman, P. Mourguiart, Quat. Res. 43, 344(1995).29. M. B. Abbott, M. W. Binford, M. Brenner, K. R. Kelts,Quat. Res. 47, 169 (1997).30. S. L. Cross, P. A. Baker, G. O. Seltzer, S. C. Fritz, R. B.Dunbar, Holocene 10, 21 (2000).31. L. G. Thompson et al., Science 282, 1858 (1998).32. A. Berger, P. J. Loutre, Quat. Sci. Rev. 10, 297 (1991).33. We thank the following: F. A. Street-Perrott andthree anonymous reviewers for comments on ear-lier drafts of this report; M. Suarez Riglos and M.Saldias for use of the herbarium at the NaturalHistory Museum, Santa Cruz, Bolivia; J. Ratter foruse of the herbarium at the Royal Botanic GardenEdinburgh; R. Guillen, L. Sanchez, M. Siles, J. Surubi,and P. Soliz for assistance with fieldwork; “Funda-cio´n Amigos de la Naturaleza” for permission towork in NKMNP; and R. Pollington for cartographicassistance. This work was funded by a Royal Soci-ety research grant (F.E.M.), two Royal Society over-seas study visits (F.E.M.), a University of Leicesterresearch grant (F.E.M.), and a University of Leices-ter Ph.D. studentship (R.B.). Funding for radiocar-bon dates was provided by the Natural Environ-ment Research Council (26 dates) and the Univer-sity of Leicester (1 date).11 August 2000; accepted 23 November 2000R E P O R T S22 DECEMBER 2000 VOL 290 SCIENCE www.sciencemag.org2294onMarch23,2011www.sciencemag.orgDownloadedfrom